WO2013087028A1

WO2013087028A1 - Iris recognition method and iris recognition method based on multi-directional gabor and adaboost

Info

Publication number: WO2013087028A1
Application number: PCT/CN2012/086678
Authority: WO
Inventors: 王琪; 张祥德; 单成坤; 刘洋
Original assignee: 北京天诚盛业科技有限公司
Priority date: 2011-12-16
Filing date: 2012-12-14
Publication date: 2013-06-20
Also published as: CN102521575B; CN102521575A

Abstract

Disclosed are an iris recognition method and an iris recognition method based on multi-directional Gabor and Adaboost. The method includes: step 1) retrieving a 2D Gabor feature from a normalized iris image; and step 2) performing classified recognition on the feature obtained in step 1) by using an Adaboost algorithm. Multi-directional Gabor wavelets in one same dimension are adopted to retrieve the feature, at the same time the expanded iris image is divided into blocks, and in combination with the integral information and local information of the iris, the Gabor features of the entire iris image and an iris image submodule are retrieved at the same time and coding is performed. Next, the integral information and the local information are combined to form a multidimensional feature vector, and the Adaboost algorithm is introduced to select features, and eventually a classifier is constructed for recognition. The beneficial effects of the present invention lie in that: only a part, containing a few noises, of the iris is used for recognition, thereby reducing the influences of noises, desirably solving the problem in recognizing a low quality iris image, and having desirable recognition performance.

Description

Iris recognition method and iris recognition method based on multi-directional Gabor and Adaboost Technical Field The present invention relates to digital image processing and pattern recognition, and more particularly to an iris recognition method and multi-directional iris recognition method.

The iris recognition methods of Gabor and Adaboost belong to the technical field of biometric identification and security authentication. BACKGROUND In modern society, with the rapid development of network technology and the rapid and frequent flow of people, a safe, reliable, convenient and efficient identity authentication system is particularly important. There are two main types of traditional identification: identifying items (keys, ID cards, etc.) and identifying knowledge (user names, passwords, etc.). However, in practical applications, it has shortcomings such as easy loss, easy forgery, non-uniqueness, and relatively small application range, which makes people urgently need an identification method that can overcome the above-mentioned shortcomings. Driven by demand, Recognition technologies based on biometrics such as face, fingerprint, iris, hand shape, handwriting, etc. have emerged. At present, there are many iris recognition systems at home and abroad. Iris feature extraction is one of the main factors affecting the performance of iris recognition systems. The existing iris Most recognition algorithms have high requirements for the quality of iris images. However, the iris images collected under natural light will be affected by the eyelashes, eyelids, light, shaking, etc., resulting in poor iris image quality. Quality iris images require an effective iris feature extraction and recognition algorithm. SUMMARY OF THE INVENTION The purpose of the present invention is to provide an iris recognition method, especially an iris recognition method based on multi-directional Gabor and Adaboost, which has good recognition performance This overcomes the above-mentioned shortcomings of the prior art. The purpose of the present invention is achieved through the following technical solutions: A multi-directional Gabor and Adaboost iris recognition method, which includes the following steps:

1) Extracting two-dimensional Gabor features from a normalized iris image, which specifically includes the following steps:

1.1) Divide the expanded iris image evenly into M rows and N columns to obtain M×N iris image sub-modules;

Replacement page (Article 26 of the Rules) 1.2) Use Gabor filters with eight directions of the same scale to act on step 1.1) The normalized iris image sub-module, and then encode the positive and negative results of the image filtering according to the Gabor real part; the expression of the Gabor filter The formula is as follows:

(j{x, y) = exp ^{1 1} exp, x = xcos9 + ysm0, χ>

y'= -xsm0 + ycose, is the direction of the Gabor filter, 6» = z'r/8, = 0,1,2,···7, [/ and are the horizontal and vertical center frequencies of the Gabor filter, respectively , Δ and are the spatial constants of the Gaussian envelope along the X axis and the axis respectively, which represent the scale of the Gabor filter; the expression of feature encoding is as follows:

^C {R _e .Im} y) = ^S g ⁿ {Re.Imi U ^, * ^G ( ^X ^ ]; its 屮/is the iris image, G is the Gabor filter, sgn ( _Relmi represents the real part of the filtering result, c _{ReIm> is the feature code:

1.3) According to the formula:

HQ

maskA Π maskB Calculate the respective Hamming distances of the MXN iris image submodules corresponding to the two iris images to be matched. For each iris image submodule, 8 Hamming distances can be obtained, and the 8 Hamming distances are composed of features The vector is denoted as V _whole , V _M = [D _Khalel 'HD _HhoIe2 ,···, HD _Klwk ]; where code A and code B respectively represent the Gabor feature codes of two iris images; mask A and mask B represent two respectively The noise template of the iris image. When the value is "1", it represents the effective iris part, and when it is "0", it represents the noise; 1.4) Let L=MXN, and obtain 8XL Hamming distances V _{part according to the Hamming calculation formula in step 1.3)} , Denoted as:

V ^HD ―,, HD, — ₂ ,"' Dy .. ,,"'

1.5) Combine the Hamming distance V _{part of the} entire iris image with the Hamming distance ^ ^ of the iris image submodule to form a feature vector V with a dimension of 8+8XMXN, V=[V _who i _e ,V _part ], which is:

Replacement page (Article 26 of the Rules) ^V = i ^HD , ,'h. , …, ^HD ,, h, ^HD pan\-, …, ^HD par, L \, …, ^HD partL% 1;

2) Use the Adaboost algorithm to classify and recognize the features obtained in step 1), which specifically includes the following steps:

2.1) Set up a sample training set {(X,, ),···, (½, )}, X, = {¾,···,¾»},, e{+l,-l}. I= l, 2, ..., N, M are the dimensions of the vector jc,.

X is the multi-dimensional feature vector formed by combining the feature vector formed by the entire iris image and the feature vector after segmentation. When the two iris image sub-modules are from the same human eye, y=l, otherwise y=-l _;

2.2) By formula:

, Perform weight initialization on the sample training set, and then train and learn the weight distribution of the sample training set to obtain a weak classifier

Λ,: →{-1,1}; t=l, 2, ..., T, Τ is the number of iterations;

2.3) Select a feature in step 1), and then use the formula-=∑,C ,), ^m=1 , ² '...' ^Μ 'to calculate the weak classifier, the classification error rate in the sample training set ^, mark s , =mi _nm , when f,≥0.5, let T=t-1, and jump out of the loop;

2.4) Through the formula: α ₍ = 0.5 xln[(l, calculate the weight of the weak classifier ",;

2.5) By formula:

D _l+l (i) = D _t (i) _e xp[-a,y,h, ,)]/Z _l 'Update the weights of the sample training set, and normalize the sample training set, Among them, A(0 is the weight coefficient of the first sample in the training time, and is the normalized idler;

2.6) Finally according to the formula:

Replacement page (Article 26 of the Rules) H(x) = sign(^a,h,(^)> Obtain the final classifier, the feature of the classifier is the feature of the iris image 屮 feature vector V, and the recognition is completed. According to another aspect of the present invention, it is also proposed An iris recognition method is provided. The iris recognition method includes the following steps: step 1) normalize the iris image, and extract the two-dimensional Gabor feature of the normalized iris image; and step 2) calculate two iris images Corresponding to the Hamming distance between Gabor features, use the Adaboost algorithm to classify and recognize the two-dimensional Gabor features using the calculation result as a sample. Further, step 1) includes: performing Gabor filtering on the normalized iris image; encoding the filtered iris image according to the real part of the Gabor filter to obtain the feature encoding of the normalized iris image, and feature encoding As the initial two-dimensional Gabor feature of the normalized iris image. Further, step 1) also includes: dividing the normalized iris image into regions to obtain multiple iris image sub-modules; performing Gabor filtering on the multiple iris image sub-modules respectively; according to the real part of the Gabor filter The filtered multiple iris image sub-modules are respectively coded to obtain feature codes of the multiple iris image sub-modules, and the feature codes of the multiple iris image sub-modules are used as the initial two-dimensional Gabor features of the normalized iris image. Further, the above-mentioned regional division of the normalized iris image includes: dividing the normalized iris image into M rows and N columns, and dividing according to the rows and columns to obtain MXN iris image sub-modules. Further, in step 1), performing Gabor filtering on multiple iris image sub-modules respectively includes: filtering multiple iris image sub-modules using Gabor filters with 8 directions at the same scale, where the table of Gabor filters

The formula, x'-cosS + jvsin6>, =-xsin0 + ycos>, 6> is the direction of the Gabor filter, 6> = /;r/8, ι = 0,1, 2,·-·7, U And V are the horizontal and vertical center frequencies of the Gabor filter, respectively, and ^ and are the spatial constants of the Gaussian envelope along the X axis and the V axis, respectively. Further, separately encoding the filtered multiple iris image sub-modules according to the real part of the Gabor filter includes: respectively encoding the filtered iris image sub-modules according to the real part of the Gabor filter using the following formula: c(Re ,in,} y) = sgn _{{Re lm}} [/(, y) * G(x, y)];

Replacement page (Article 26 of the Rules) In the formula, I(x, y) is the expression of the iris image, G(x, y) is the expression of the Gabor filter, sgn _(ReIni) represents the real part of the filtering result, and c _(Relm) is the feature code. Further, calculating the Hamming distance between the Gabor features corresponding to two iris images includes-according to the formula: HD ^ ^co< ^ ^e ^ ® ^coc ^ ^e ^ Π maskA Π

maskA Π maskB calculates the Hamming distance between the Gabor features of two iris images. In the formula, codeA and codeB respectively represent the corresponding feature codes of the two iris images; maskA and maskB represent the noise templates of the two iris images respectively. Further, in step 2), using the calculation result as a sample to use the Adaboost algorithm to classify and recognize the two-dimensional Gabor features includes: constructing a vector using the calculation result of the Hamming distance between the Gabor features corresponding to the two iris images, the The vector is used as a sample to use the Adaboost algorithm for classification and recognition. Further, step 2) Use the Adaboost algorithm to classify and recognize the vector as a sample, including: Step 2.1) Set a sample training set {(Χ,,) .·,^^)}, A ={½,· . ··,}, e {+ l, - 1}, i = l, 2, ..., N, M is a vector; C rear, the dimension of which Cao, ^x is the normalized two iris images The value of y is: When the two iris image sub-modules are from the same human eye, ^=1, otherwise y = Step 2.2) Pass the formula: D, {T} = \lN,i = \,'''N, initialize the weight of the sample training set, and then train and learn the weight distribution of the sample training set to obtain a weak classifier _; h _t :x^{-\,\) t^l, 2, ..., T, where Τ is the number of iterations; Step 2.3) Select a two-dimensional Gabor feature in Step 1), and then use the formula: =∑ _Λ , ₍ , _m =l, 2, …, M, Calculate the classification error rate of the weak classifier/in the sample training set, record ε _t =min^, when _£ ,≥0.5, let T=t-1, and jump out of the loop; Step 2.4) Pass the formula: a, = 0.5xln[(l_s)/f,], calculate the weight of the weak classifier _;

Replacement page (Article 26 of the Rules) Step 2.5) Through the formula: ί, ₊ ) = £), (^ _Χ ρ[- _α /ί ₍ (χ, )]/Ζ,, update the weights of the sample training set, and normalize the sample training set, in Equation, where Dt(i) is the weight coefficient of the i-th sample in the t-th training, and is the normalization factor; Step 2.6) According to the formula: H(x) = ^«( «»)), obtain For the final classifier, the feature H(x) of the classifier is the final classifier, and the training is completed; Step 2.7) Use the Gabor code corresponding to the classifier trained by Adaboost as the final iris feature for iris image comparison. The beneficial effects of the present invention are as follows: only a part of the iris containing less noise is used for recognition, the influence of noise is reduced, the recognition problem of low-quality iris images is well solved, and the recognition performance is very good. The feature extraction of the iris image uses multi-directional Gabor wavelet at the same scale, and the expanded iris image is divided into blocks, combined with the overall and local information of the iris, and the Gabor features of the entire iris image and the iris image sub-module are extracted and coded at the same time. Then the whole and the part are combined to form a multi-dimensional feature vector, and the Adaboost algorithm is introduced for feature selection, and finally a classifier is constructed for recognition. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments and descriptions of the present invention are used to explain the present invention, and do not constitute an improper limitation of the present invention. In the accompanying drawings-Figure 1 is a flowchart of a method for iris recognition based on multi-directional Gabor and Adaboost according to an embodiment of the present invention; Figures 2 (a) to 2 (h) are 8 directions of Gabor in an embodiment of the present invention The schematic diagram of the real part filter; Fig. 3 (a) to Fig. 3 (h) are the coding process diagrams of Gabor according to the embodiment of the present invention; Fig. 4 (a) to Fig. 4 (d) are the embodiment of the present invention Schematic diagram of the block. DETAILED DESCRIPTION As shown in FIG. 1, the iris recognition method described in this embodiment is based on the multi-directional Gabor and Adaboost algorithm, and includes the following steps: Step 1) Normalize the iris image, and extract the normalization- Two-dimensional Gabor features of the transformed iris image; and

Replacement page (Article 26 of the Rules) Step 2) Calculate the Hamming distance between the Gabor features corresponding to the two iris images, and use the Adaboost algorithm to classify and recognize the two-dimensional Gabor features using the calculation result as a sample. First, step 1) Extract two-dimensional Gabor features from the normalized iris image. The 2D Gabor filter has good resolution capabilities in the time domain and frequency domain, and the expression is as follows:

2πδ _χ δ ₎ . Among them, 0 is the angle corresponding to the direction of the pixel point (x, y), that is, the direction of the Gabor filter, and the angle between the x'axis and the X axis is 0. Among them, when 0 = *^, ί'= Ο,1,2...,7, the eight-direction filter is shown in Figure 2 (a) to

8

As shown in Figure 2 (h), and are the spatial constants of the Gaussian envelope along the X-axis and the y-axis, respectively, and the value of the sum is preferably 4.0, and f is the ridge frequency. As shown in FIG. 2, the embodiment of the present invention uses Gabor filters with the same scale and multiple directions. Specifically, Gabor filters with 8 directions can be used, so that the 8 directions respectively correspond to 0 = /8, !'= 0, 1,2,...7. Apply these 8 two-dimensional Gabor filters to the normalized iris image, and use the quadrant of the filtering result to encode, which is expressed as follows: c _iReMi (y) = sgn _{{Rc lm)} [I(x, y) * G (x, y)] (2) The I is the iris image, G is the Gabor filter, sgn _|Relmi is the real part of the filtering result, and c _(ReIm> is the feature code. The entire iris area contains noise such as light spots and eyelashes. Therefore, when only a certain part of the iris with less noise is used for recognition, the influence of noise can be reduced to a certain extent, and the recognition performance may be improved. Therefore, the present embodiment can uniformly expand the iris image. Divide into M rows and N columns, so that Λ x N sub-blocks are obtained. Figures 4 (a) to 4 (d) are schematic diagrams of iris block in an embodiment of the present invention, in which Figure 4 (a) and Figure 4 (b) ) Is the iris image to be divided into 3 rows and 4 columns according to the annular area of the iris image (ie M=3, N=4), so that 12 sub-blocks are obtained. In the figure, R1, R2, R3, Ul, U2, U3> Ll, L2, L3, Dl, D2, and D3 are the 12 sub-blocks obtained after block division. In order to facilitate Gabor coding, the fan-shaped area of the original iris image is mapped to a square area using biquadratic linear interpolation, as shown in Figure 4 (c ) And Figure 4 (d) are the obtained MXN=12 sub-blocks used for Gabor coding. Thus, in addition to the overall recognition, different parts of the image are recognized separately at the same time to improve the accuracy of recognition.

Replacement page (Article 26 of the Rules) According to the division method shown in Fig. 4(a) to Fig. 4(d), the above-mentioned specific method of dividing the annular area of the iris image into 3 rows and 4 columns can be: taking the center of the iris as the center of the circle, and making circles according to different radii to divide It is 3 circular rings, divided into 4 radians evenly according to the radians, thereby obtaining 12 regional blocks for interpolation mapping. In step 2) of the present invention, the Hamming distance is used as the similarity measure, and the Gabor codes of the two images are XORed. As shown in Figure 3, Figure 3 (a) and Figure 3 (b) are schematic diagrams of the real and imaginary parts of the Gabor filter, respectively, and Figure 3 (c) and Figure 3 (d) are two normalized iris images, Figure 3 (e) and Figure 3 (0 is the first iris image Figure 3 (c) Gabor code filtered by the Gabor filter of Figure 3 (a) and Figure 3 (b), Figure 3 (g) and Figure 3 (h) is the second iris image. Figure 3 (d) and the Gabor code obtained by the Gabor filter in Figure 3 (a) and Figure 3 (b). Figure 3 (e) ) And Figure 3 (g) for XOR operation, Figure 3 (f) and Figure 3 (h) for XOR operation. That is, when matching, compare Figure 3 (e) and Figure 3 (g), Figure 3 (f) Compare with Figure 3 (h). Consider the noise of two images, add the XOR result of effective pixels and divide by the number of effective pixels as a feature value, which is the Hamming distance. The formula is as follows:

Its codeA and codeB are the Gabor feature codes of iris image A and iris image B respectively; and respectively represent the noise template of iris image A and iris image B. When the value is "1", it represents the effective iris part, and when it is "0" Represents noise. First, calculate the Hamming distance of the Gabor encoding of the entire iris expanded image. Since the present invention selects Gabor filters in 8 directions, a vector composed of 8 Hamming distances can be obtained, which is denoted as: V _whole = [HD _KhM , HD _Me2 , ..., HD _Khokt ] (4) Then, calculate the Hamming distance corresponding to each of the blocks corresponding to the two iris images. For each sub-block, 8 Hamming distances can be obtained correspondingly, let = ATxN, You can get 8>< Hamming distance, which is recorded as:

= iHD —,, HD _ ₂ , --, HD — _S ,"', HD _L ―,,... 'U (5) Finally,'Combine ^ with ^,,' to get the vector = ,. _te , ^„ ], that is: = [^ _Ao , _fl,, . _fe8 , ^—,— ,HZ), ---,HD — ₈ ] (6)

Replacement page (Article 26 of the Rules) In this way, a vector composed of the Hamming distance of the two iris images after the two iris normalization is obtained, and the dimension is: 8 + 8xMxN.

The Adaboost algorithm uses a large number of simple classifiers (weak classifiers) with general classification capabilities, which are combined through certain methods to form a classifier with strong classification capabilities. The specific process of using Adaboost algorithm to classify and recognize the above vectors is as follows: Input: training set {(w' U ，其屮X, = {,···,} 'y, e{+l,-\}, i =, \,l -'N, M is the dimension of the vector c,.; the number of iterations T and the weak learning algorithm. Initialization: weight!) ): ^ ,/:^... (7) Operation: For ζ = 0,1,2,·--Γ 1) Learn from the training set with weight distribution to obtain a weak classifier

¾,:χ→{-1,1} (8)

2) Choosing the best feature to minimize the classification error rate ^, its 屮

^£ _m =∑ M^y ^W ^ (9) m = 0,1,2,……M, and denote =min£ _m . If ≥0.5, let Γ = ί_1 and jump out of the loop. 3) Calculate the weight of the weak classifier:

4) Update sample weight: +ι (0 = A ('·) exp[- {x _t (11) where 4 is the normalization factor. Output: H(;c) = g"(∑ a,A,( x)).

Replacement page (Article 26 of the Rules) The iris recognition problem is a typical classification problem. As long as each weak classifier (the Hamming distance calculated above) corresponds to 1 Gabor code, and the classification judgment is made according to the size of the weak classifier threshold, the iris recognition process is the process of Adaboost selecting features, that is The process of selecting weak classifiers. In the process of classifying two images, use V = [HD' _hoM ,..., HD, HD^ ^,-, HZ) — ₈ ] (12) as a sample, that is, _{X in the} training set; shooting for the same human eye The two images obtained, let = 1, and the two images taken by different human eyes, let ^ = -1. Then use the Adaboost algorithm for training, select good classification features, and then combine these classification features into a stronger classifier. Combine the classifiers trained by the above-mentioned Adaboost algorithm, and each classifier corresponds to the Gabor code of a specific sub-block. The Gabor code is the final Gabor feature of the iris, which is used for iris comparison. Using the method of the above embodiment, the feature extraction of the iris image adopts multi-directional Gabor wavelets at the same scale, and at the same time, the expanded iris image is divided into blocks, combined with the overall and local information of the iris, and the entire iris image and the iris image are extracted at the same time. The Gabor feature of the module is coded, and then the whole and the part are combined to form a multi-dimensional feature vector, and the Adaboost algorithm is introduced for feature selection, and finally a classifier is constructed for recognition. Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general computing device, and they can be concentrated on a single computing device or distributed in a network composed of multiple computing devices. Above, optionally, they can be implemented by program codes executable by the computing device, so that they can be stored in the storage device and executed by the computing device, or they can be made into individual integrated circuit modules respectively, or they can be made into various integrated circuit modules. The multiple modules or steps are made into a single integrated circuit module to achieve. In this way, the present invention is not limited to any specific combination of hardware and software. The above are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Replacement page (Article 26 of the Rules)

Claims

Claims, a method for iris recognition based on multi-directional Gabor and Adaboost, characterized in that it includes the following steps: 1) extracting two-dimensional Gabor features from the normalized iris image; and 2) making the ffl Adaboost algorithm to perform step 1 ) To classify and recognize the features obtained. The method for iris recognition based on multi-directional Gabor and Adaboost according to claim 1, characterized in that, step 1) specifically includes the following steps:

1.1) Divide the expanded iris image evenly into M rows and N columns to obtain MXN iris image sub-modules;

1.2) Use Gabor filters with the same scale and eight directions as the iris image sub-module in step 1.1), and then encode the positive and negative results of the image filtering according to the Gabor real part.

The expression of Gabor filter is as follows:

θ, y'^-xsm9 + ycos0,> is the direction of the Gabor filter, Θ = ίπ1%, ί = 0, 1, 2.·7, ί/ are the horizontal and vertical center frequencies of the Gabor filter, δ _χ and are the spatial constants of the Gaussian envelope along the X axis and the _y axis, respectively, which represent the scale of the Gabor filter;

The expression of feature encoding is as follows:

im} 0, y) = sgn _{Re , _Im , [/ (x, y) * G(x, y)]; where / is the iris image, G is the Gabor filter, and sgn^^ represents the real and virtual of the filtering result Department symbol, which is the feature code;

1.3) According to the formula:

maskA Π maskB

Replacement page (Article 26 of the Rules) Calculate the respective Hamming distances of the MXN iris image submodules corresponding to the two iris images to be matched. For each iris image submodule, 8 Hamming distances can be obtained, and the feature vector composed of the 8 Hamming distances is recorded as V _who ie, V _Khole = [HD

Among them, code A and code B respectively represent the Gabor feature codes of two iris images; mask A and mask B represent the noise templates of two iris images respectively, the value of "Γ represents the effective iris part, and the value of "0" represents the noise ；

1.4) Let L=MXN, follow the step 1.3) Hamming calculation formula to get 8XL Hamming distance V _par …… Marked as: ',''',,,,:

_{1.5) Combine the Hamming distance {} ^ of the entire iris image and the Hamming distance V _{whole of the} iris image submodule to form a feature vector V with a dimension of 8+8XMXN, V=[V _wh0 , _e ,V _part ] , which is:

V = [D _thoM ,..., HD _M , HD, ■■-, HD^ ,.■·, HD一].

The method for iris recognition based on multi-directional Gabor and Adaboost according to claim 2, characterized in that step 2) specifically includes the following steps:

2.1) Set up a sample training set {(^ ,),···, V)}, _Xl ={x ,-,x _lM }-j _{l G} {+l,-l}, i=l, 2, ..., N, M are the dimensions of the vector χ,., where: X is the multi-dimensional feature vector formed by combining the feature vector formed by the entire iris image and the feature vector after the block, when the two iris image sub-modules come from For the same person, y = l, otherwise-1;

2.2) Through the formula: D\{i, =

\,-"N, initialize the weight of the sample training set, and then train and learn the weight distribution of the sample training set to obtain-· weak classifiers; i _i;

^ _:;C→ {-1,1} _{; t} =l, 2, ..., T, Τ is the number of iterations;

2.3) Select the characteristics of the step 1) 屮, and then use the formula:

Calculate the weak classifier /ί, the classification error rate in the sample training set, record =1^11^, when ≥0.5, let T=t-1, and jump out of the loop;

2.4) by the equation: ct, = 0.5χΙη [(1_ίΓ ,) / £,], weak classifiers calculate the weight [alpha] _([alpha] re _(;

Replacement page (Article 26 of the Rules)

2.5) By formula: £», ₊₁ (0

, Update the weight of the sample training set, and normalize the sample training set, where is the weight coefficient of the first sample in the Zth training, yes! )υ ·'Chemical factor;

2.6) According to the formula: H (; c) =

, Obtain the final classifier, the feature of the classifier is the feature of the feature vector V of the iris image, and the recognition is completed. 1. An iris recognition method, characterized in that it comprises:

Step 1) Normalize the iris image, and extract the two-dimensional Gabor features of the normalized iris image; and

Step 2) Calculate the Hamming distance between the Gabor features corresponding to the two iris images, and use the calculation result as a sample to classify and recognize the two-dimensional Gabor features using the Adaboost algorithm. 3. The iris recognition method according to claim 4, wherein the step 1) comprises:

Gabor filtering is performed on the normalized iris image;

The filtered iris image is coded according to the real part of the Gabor filter to obtain the feature code of the normalized iris image, and the feature code is used as the initial two-dimensional Gabor feature of the normalized iris image . The iris recognition method according to claim 5, wherein the step 1) further comprises: dividing the normalized iris image to obtain a plurality of iris image sub-modules; The image sub-modules respectively perform Gabor filtering;

The filtered multiple iris image sub-modules are respectively coded according to the real part of the Gabor filter to obtain feature codes of the multiple iris image sub-modules, and the feature codes of the multiple iris image sub-modules are taken as The initial two-dimensional Gabor feature of the normalized iris image. The iris recognition method according to claim 6, characterized in that, dividing the normalized iris image into regions comprises-dividing the normalized iris image into M rows and N columns,

Divide according to the ranks to obtain M×N iris image sub-modules. The iris recognition method according to claim 6, characterized in that step 1) performing Gabor filtering on the plurality of iris image sub-modules respectively comprises:

Replacement page (Article 26 of the Rules) The multiple iris image sub-modules described by the Gabor filter with 8 directions of the same scale are used for filtering. The expression of the Gabor filter is as follows:

Equation, X'= cos ^l + _ysin6*, y'= -xsinB + ycosB, (9 is the direction of the Gabor filter, θ = ϊπ, ί· = 0, 1, 2, ··7, C7 and Gabor respectively The horizontal and vertical center frequencies of the filter, and are the spatial constants of the Gaussian envelope along the X axis and the axis, respectively. The iris recognition method according to claim 6, characterized in that, according to the real part of the Gabor filter Encoding the multiple filtered iris image sub-modules respectively includes:

Use the following formula to code the filtered iris image sub-modules according to the real part of the Gabor filter:

， = Sgn [I(x,y)*G(x,y)]; In the formula, I (x, y) is the expression of the iris image, and G (x, y) is the expression of the Gabor filter' sgn^ _lm| represents the real part of the filtering result, and c^.w is the feature code. The iris recognition method according to claim 4, wherein calculating the Hamming distance between the Gabor features corresponding to two iris images comprises:

According to the formula:

Calculate the Hamming distance between the Gabor features of two iris images. In the formula, code A and code B respectively represent the corresponding feature codes of the two iris images; mask A and mask B represent the two irises respectively The noise template of the image. 1. The iris recognition method according to claim 4, characterized in that, in the step 2), using the calculation result as a sample and using the Adaboost algorithm to classify and recognize the two-dimensional Gabor feature comprises: using two iris images corresponding to Gabor The calculation result of the Hamming distance between the features constructs a vector, which is used as a sample for classification and recognition using the Adaboost algorithm. The iris recognition method according to claim 11, wherein the step 2) using the Adaboost algorithm to classify and recognize the vector as a sample comprises:

Replacement page (Article 26 of the Rules) Step 2.1) Set up a sample training set {( _½ ),..., ( _½ , ^)}, X, = {X,,,···, ¾},

^ ei+1,-1}, i=l, 2, ..., N, M is the dimension of the vector χ,., where is the vector composed of the Hamming distance of the two iris images after normalization, The value of y is: When the two iris image sub-modules are from the same human eye,; = 1, otherwise: v = -l; Step 2.2) Pass the formula: D\{i,

Λ,: →{-1,1} _; t=l, 2, …, T, Τ is the number of iterations; Step 2.3) Select a two-dimensional Gabor feature in Step 1), and then use the formula: ^ ( ^x ') , m=l, 2, …, M, calculate the classification error rate of the weak classifier/in the sample training set^, mark = ηώ! , When ≥ 0.5 inch, let T=t-1, and jump out of the loop; Step 2.4) Pass the formula: α ₍ = 0.5 xln[(l- calculate the weight of the weak classifier ",; Step 2.5) Pass the formula: β, ₊₁ (0 = £>,(ζ χρ[-α^Λ( )]/ ^ζ ₍ , update the weights of the sample training set, and normalize the sample training set, in the formula, where (0 is the first ί' The weight coefficient of each sample in the first training, Z, is the normalization factor; Step 2.6) According to the formula: H(x) = ^w(iZ,A,(x)), the final classifier is obtained, and the classification The feature Η (X) of the detector is the final classifier, and the training is completed;

Step 2.7) Use the Gabor code corresponding to the classifier trained by Adaboost as the final iris feature for comparison of iris images.

Replacement page (Article 26 of the Rules)